01979 by: Schweizerische Chemische Gesellschaft, Basel,

Sociktk suisse de chimie, BUe - Societh svizzera di chimica, Basilea

ISSN 0018-019X
Nachdruck verboten - Tous droits rkservts - Printed by Birkhguser AG, Basel, Switzerland
Erscheint 9mal jahrlich - Parait 9 fois par an

177. Kinetics of Ethylene Oxidation on a Supported Silver Catalyst

by Hans Rudolf Dettwiler, Alfons Baiker and Werner Rieharz
Technisch-Chemisches Laboratorium, Eidgenossische Technische Hochschule, CH-8092 Zurich

(17.1V.79)

Summary
The kinetics of ethylene oxidation by air over a supported silver catalyst were
investigated in the temperature range 490-620 K. The reaction network was found
to be triangular. Under virtually constant oxygen partial pressure (0.2 bar), the
following rate relationships, (in mol gels-') were found: (formulae in curved brack-
ets denote partial pressures)

r3= k3 (C2H40) +
C2H4O 512 0 2 k3 2C02+2H20

where: ,]:-9:
k,=40.69 exp [ ~ [mol . g-'. s-'] T > 490 K ;

[mol . g-I . s-'1 T > 490 K ;

k3= 9.51. 1O3exp[-=], [mol . g-' . s-I . bar-'] T > 530 K ;

KC2Hq= 1.84. lop4e x p [ E ] , [bar-']

R is expressed in J . mol-' . K-'.

The given rate expressions are discussed in the framework of earlier kinetic in-
vestigations.

oO18-019X/79/6/ 1689- 12$01.00/0 0 1979 SchweizerischeChemische Gesellschaft

1690 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979) - Nr. I77
CHIMICA

Introduction. - Since Lefort's original description of the prlocess [l], the oxida-
tion of ethylene has been the subject of numerous investigations [2] [3]. The re-
ported kinetic equations extend from empirical correlations to complicated rate
equations, mostly of the Langmuir-Hinshelwood type. Some of these equations
account for the observed inhibiting effect of reaction products. Differences in the
reported kinetic equations and mechanism are primarily due to opposing opinions
about the type of chemisorbed oxygen present at the catalyst surface and to the
different kinds of supported or unsupported silver catalysts which were used for
the investigations. The result of these various individual efforts is therefore not a
generally accepted rate equation but a variety of relationships, valid under the
conditions reported. In the present work, we studied the kinetics of ethylene oxida-
tion over a supported silver catalyst in the temperature range 490-620 K and in
the presence of excess oxygen.
Experimental. - Apparatus. The apparatus (Fig.l ) , which was used for the kinetic measurements,
consisted of a gas metering system, a Pyrex U-tube reactor ( i d . 1 cm) and a gars chromatograph (Carlo
Erba, Mod. Fractovap 2200) equipped with a HW-detector and a sample valve. The reactor tube was
thermostated by a temperature controlled silicon oil bath.
Analysis. The GC. analysis was carried out by means of a Porapak S Column (1 m x 4 mm).
Column temperature was 100" and helium flow rate 0.68 cm:s--l. Calibration was carried out with
each species.
Catalyst. This was prepared by impregnation of pumice particles of 0.75 mm average diameter:
60 g lactic acid (90%) was heated to 80" on a water bath. 20 g silver oxide (.Fluka A G ) and 0.5 cm3
H 2 0 2 (30%) were then added in small portions to the continuously stirred solution. After the silver
oxide had dissolved completely the solution was diluted with 50 cm3 distilled water. The impregna-
tion was carried out with the above solution in a rotating evaporator containing 200 g of pumice. After
impregnation the excess liquid was removed and the particles were cooled. To decompose the silver
lactate the catalyst was then slowly heated to 400" under nitrogen. After 20 h the catalyst was removed
from the furnace, cooled in a nitrogen flow and weighed. N o further treatment was given to the catalyst
before its use in the reactor. The catalyst had the following properties: silver content, ca. 8%; specific
BET-surface, 1.6-2 m2 8-l; specific pore-volume, 0.452 cm3 gg'; mean port: diameter, 7 . lo3 nm;

Fig. I. Apparatus
1 low flow valves
2 flow meters
3 manometer
4 drying unit
5 reactor
6 thermocouple well
7 oil bath
8 sampling valve
9 gas chromatograph
 HELVETICA
CHIMICA
ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177 1691

xE:s

0.5 ?
0.3

O1 t Fig.2. Long lerm behaviour

T= 573 K
(C2H4)=0.03bar
(02) = 0.2 bar

apparent density, 1.18 g ~ m - ~ ; density, 1.68 g ~ m - Mean

true ~ . pore diameter and specific pore volume
have been determined by the mercury penetration method.
Procedure. Before starting with the experimental runs, the reactor temperature was adjusted to a
given value. Gas samples were taken approximately 20 to 30 min after setting the desired flow rates.
The mol fractions of ethylene oxide and carbon dioxide in the product stream were obtained from
the chromatogram with the aid of calibration curves. The concentration of water was assumed to be
equal to the measured concentration of carbon dioxide. Average reaction rates were calculated, based
on the assumption of a differential reactor (conversion< 10%). The calculated reaction rates are there-
fore assumed to correspond to the average partial pressures of reactants and products in the reactor.
Temperature gradients were small enough to be neglected.

Results. - Long term behaviour, mass transfer and reaction network. The long
term behaviour of the catalyst was tested during preliminary runs. Figure 2 shows
the declining activity and selectivity observed during these experiments. The
kinetic measurements were carried out in the first min, where the activity and
selectivity could be assumed to be approximately constant (loss of activity less
than 3%).
Figure3 shows the measured selectivity as a function of gas flow rate and
temperature at constant contact time. The selectivity increases with increasing gas
flow rates, but with flow rates greater than 45 cm?s-l (Red') 2 10.5) the influence
becomes very small. In order to avoid the disguise of the kinetic measurements
by gas flow rate influences, all experiments were carried out with a gas flow rate
of 45 cm;s-l. The absence of mass transfer limitation of the reaction under the
conditions of the kinetic experiments was assured by applying the test criterion of
Mears [4]. The estimated Damkohler number amounted to Da=0.053 at 670 K
and indicated that the reaction was not limited by mass transfer (DaQ 1).
I) Red is the Reynolds number referred to the catalyst particle diameter: Red= m' d/p, where
m* = mass flow rate per unit cross-sectional area, d = catalyst-particle diameter and p = dynamic
viscosity.
1692 HELVETICA
CHIMICA
ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177

Figure 4 showing selectivity versus conversion, substantiates that with the cata-
lyst used the following reaction network has to be considered.

rl /"\
+CHZ-CH~

Kinetics. For the quantitative kinetic analysis, only those experimental runs
where ethylene conversion was less than 10% were used for the determination of
the reaction rates. The differential reactor experiments were evaluated by means
1 AN, 1 1
of the following relationship: rl= - - . -- ANY = - . N: . X, where
w At W W
r, reaction rate (mol s-lg-'), w catalyst weight (g), NT molar flow rate (mol s-I),
Xi conversion (- ).
Figure 5 shows the measured rate of ethylene consumption as a function of the
ethylene partial pressure at different temperatures and under virtually constant
oxygen partial pressure (0.2 bar). A first order dependence of the ethylene reac-
tion rate on oxygen partial pressure in the temperature range 493-623 K is indi-
cated in Figure 6.

S
t
0.5 -

0.3 -

0.1 -

473 523 573 623

Fig.3. Selectivity versus gasflow rate and temperature at constant contact time t = V f v".
V = volume of fixed bed, V* = gas flow rate; (CzH4) = 0.03 bar; ( 0 2 ) = 0.2 bar
 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177
CHIMICA 1693

S
I0

0.5

Fig.4. Selectivity versus conversion

T = 523 K (--- expected dependence of
S on X, in the case of pure consecutive
XE reaction)

/A
/O- -0-0-
533

"
"493 K

0
I
0,Ol
I I
0.02
1 I
0.03
,p, ( bar )

Fig.5. Measured rate of total ethylene consumption as a function of ethylene partial pressure and tem-
perature

The oxidation of ethylene oxide was studied separately and Figure 7 shows
the determined dependence of the rate on the ethylene oxide partial pressure. In
the temperature range 533-623 K and under the given partial pressure conditions
1694 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979)
CHIMICA - Nr. 177

(excess oxygen), the rate r3 was first order with respect to the ethylene oxide partial
pressure.
r3= k , (C2H40). (1)

(c2H40)
The measured selectivity S = - for the oxidation of ethylene to
(c2H4)0- (C2H4)
ethylene oxide as a function of temperature is shown in Figure 8. The reaction rates
rl and r2 were derived assuming that the ethylene oxid formed by r l reacts further
with rate r3 and utilizing the measured selectivity S (cf: Figure 8) as well as the re-
action rate obtained for total ethylene consumption rE.
The determined reaction rates r l and r2 are in accordance with the following
relationships:

r- 1,7s

2 -
- V*-

P -
%H4
45Cm3IS

0,022-0,024 bar

0’
/
-

1 -
/./
Fig.6. Dependence of the rate of total ethylene consumption on ihe oxygen portialpressure
 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177
CHIMICA 1695

r3 .lo6( mol.g! 61)

10 ’

0 0.01 a02
Fig.1. Rate of ethylene oxide oxydation as a function of ethylene oxide partial pressure and temperature

Fig.8. Dependence of
Selectivity S on reaction
temperature at low
470 520 5 70 conversion (XE< 0.1)
1696 HELVETICA ACTA- Vol. 62, F a x . 6 (1979)
CHIMICA - Par. 177

where quantities between curved brackets denote partial pressures and KC2H4
(bar-') is the adsorption equilibrium constant of ethylene. The rate equations (2)
and (3) correspond to a Langmuir-Hinshelwood expression with the assumptions:
surface reaction controls, ethylene and oxygen are competing for different sites on
the surface and oxygen is only weakly adsorbed.

With 1S (02)Ko2equations (2) and (3) are obtained.

The assumption of ethylene and oxygen having affinity to different sites on
the surface is we11 documented [ll]. In the case of excess oxygen, (2) and (3) can

In ki In KC2H4

-7 3

-0 1

-9 6

- 10 5
\
-1 1 4

-1 2 3

-1 3 2

-1 4 1

-15 0

-16 -1

-17 -2

1
I I I I
- 9. Temuerature
Fig.
1.6 1.7 1.8 1.9 2.0 dependence of rate constants
and adsorption equilibrium
103/T [K-'1 constafit
 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177
CHIMICA 1697

Table 1. Rate equations for rl

Rate expression Reaction conditions Activation Energy E, Ref.
r1 Heat of adsorption dqo
Ea A qo
kJ mol-I kJ mol-’

k(C2H4) (C2H4)= 0.0134-0.0536 bar

(02)=iI= 0.0402-0.1609 bar
Cat:Ag/glass wool fibers
T = 536 K

(C2H4)= 0.0204-0.18 bar

(02)air=0.21-0.176 bar
Cat:Ag(Ba)/
Carrier not specified
T = 497-547 K
(C2H4)= 0.92-0.153 bar
(02)= 0.1 1-0.866 bar
Cat: Ag( Ba)/Alumina
T = 513-533 K

k(Oz) ((02)>0.1 bar) (C2H4)= 0.1-0.8 bar

(02) = 0.2-0.6 bar
Cat: Ag(Ba)lKorund
T = 513 K

(C2H4) = 0.0016-0.054 bar

(02)air= 0.2 13-0.202 bar
Cat: Ag/Alurnina
T = 523-573 K
(C2H4)= 0.561 bar
(02)= 0.0966 bar
Cat: AgAlumina
T=448-523 K

(CzH4)= 0.026- 1.32 bar

(02)=0.061-1.32 bar
Cat: Ag/Alumina
T=493 K

(C2H4) = 0.026-0.062 bar

( 0 2 ) = 1.31-13.1 bar
Cat:Ag/Alumina
T=493 K

(C2H4) = 0.1-0.9 bar

(02)= 0.1-0.9 bar
Cat:Ag precipitated
T=494 K
(C2H4) = 0.002-0.03 bar 71 54 present
(02) = 0.05-0.2 bar work
Cat: Ag/Pumice
T = 493-623 K
1698 HELVETICA ACTA- Vol. 62, F a x . 6 (1979) - Nr. 177
CHIMICA

Table 2. Rare equationsfor i-2

______
Rate expression Reaction conditions Activation Ebergy E , Ref.
r2 Heat of adsorption Aq,

(C,H,)= 0.0134-0.056 bar 67 -

(02)alr=0.0402-0.1609 bar
Cat:Ag/glass wool fibers
T= 536 K
(C*H4)= 0.0204-0.18 bar 63
(02)alr= 0.21-0.176 bar
Cat: Ag(Ba)/Carrier not specifled
T = 497-547 K
(C2H4)=0.1-0.8 bar 138 80
(02)= 0.2-0.8 bar
Cat:Ag(Ba)/Korund
T = 533 K
(C,H4)= 0.0016-0.134 bar 31
(02)= 0.213-0.186 bar
Cat:Ag/Alumina
T = 433-493 K
(CzH4) = 0.561 bar - -
( 0 2 ) = 0.0966 bar
Cat: Ag/Alumina
T=448-523 K
(C2H4)= 0.026-1.32 bar - -
(0~)=0.061-1.32bar
Cat: Ag'Alumina
T=493 K
- k(CZH4)(02)-o.3(high(02))
~ ( C Z H ~ ) ( O Z ) ~ ' ~ ( ~ O W(C2H4)=
( ~ Z ) ) 0.026-0.062 bar
(02)= 1.3 13.1 bar
1-
- -

Cat: Ag/Alumina
T=493 K
k (C2H4)-02(02)'" (C2H4)=0.1-0.9 bar - - ~ 3 1
(02)=0.1-0.9 bar
Cat:Ag precipitated
T=494 K
Kc~Hq(~2H4) (C2H4)=0.002-0.03 bar 63 54 present
k.-
+ KC1H4(C2H4) (02)= 0.05-0.2 bar work
Cat: Ag/Pumice
T = 493-623 K

be simplified by considering the oxygen partial pressure as a cor.stant according to

ki= ki (02)leading to the equations:

where i = 1 and 2.
 HELVETICA ACTA- Vol. 62, Fasc. 6 (1979) - Nr. 177
CHIMICA 1699

From the evaluated kinetic data, we obtained the following Arrhenius relation-
ships for the three reaction rate constants: k , = 40.69 . exp [- 76 970/R . TI,
[mol.g-'. s-'1; k2=1.34.exp [-63070/R.T],[m0l.g-~.~-']; k3=9.51. 103.exp
[- 85 280/R . TI, [mol . g-' . s - I bar-']; where R is expressed in J . mol-I . K-'.
The adsorption equilibrium constant K,,, was estimated by a non linear least
square regression method and was in accordance with:

The Arrhenius plots of k,, k 2 and k, as well as a Van't Hoffplot of KC2H4are

summarized in Figure 9.
Discussion. - Tables 1-3 give a survey of the kinetic rate equations which have
been reported for rl, r2 and r3, respectively. Despite the variety of proposed rate
equations, there are some general characteristics with respect to the present work.
Several investigators (51 [6] [ 101 [ 121 reported the same partial pressure depen-
dence for both reaction rates, rl and r2. Moreover rl and r2 show apparent activa-
tion energies of almost the same magnitude [ 5 ] [6] [9]. Most investigators [ S ] [lo]

Table 3. Rate equations for rl

Rate expression Reaction conditions Activation Energy E, Ref.

r3 Heat of adsorption dqo
Ea A qo
kJ mol-I kJ mol-'
(C2H4) = 0.0134-0.0536 bar -
(OZ)=~ 0.0402-0.1609
~= bar
Cat:Ag/glass wool fibers
T=536K
k , KC2H40('2H40) (C2H4)= 0.2-0.5 bar 54
+ KC2H40(C2H40) ( 0 2 ) = 0.2-0.8 bar
Cat:Ag(Ba)/Korund
T = 533 K
(CzH4) = 0.02 bar 71-80
(O&= 0.2058 bar
Cat:Ag/Carrier not specified
T = 560 K
(C2H4)= 0.0016-0.054 bar 48
( 0 2 ) = 0.213-0.202 bar
Cat:AgAlumina
T = 523-572 K
(C2H4)= 0.0128-0.0488 bar -
(02)air=0.21 140.2046 bar
Cat: Ag/Alumina
T=473-558 K
(C2H4) = 0.005-0.025 bar 85
( 0 2 ) = 0.2 bar
Cat: Ag/Pumice
T- 533-623 K
1700 HELVETICA ACTA- Vol 62, Fasc. 6 (1979) - Nr. 177
CHIMICA

[14] found first order dependence of r3 on the ethylene oxide partial pressure. At
low ethylene oxide partial pressures the rate expression reported by Fognani &
Montarnal [8] can also be considered to be approximately first order with respect
to ethylene oxide, because the ethylene oxide is reported to be only weakly adsorbed
(qo- 2 1 kJ mol-I).
The vast majority of the reported rate equations were derived without taking
into account inhibition by the products and it is likely that some kinetic results were
affected by inhibition, especially those where the kinetics were derived from mea-
surements at higher conversion. The inhibition by the reaction products has been
studied carefully by Metcalf & Harriot [ 121. They reported a rapidly reversible in-
hibition effect by carbon dioxide and water, while the inhibition by ethylene oxide
was found to be only slowly reversible. Although there are some similar character-
istics in the reported rate equations, the derivation of the mechanisms on the basis
of these kinetic investigations proved to be questionable and speculative. Some
mechanisms which have been proposed on the basis of further information, as ad-
sorption and tracer studies are discussed in references [3]. The best word on the
question about mechanism will come only from investigations which identify the
surface intermediate directly under reaction conditions, as has been done by Force
& Bell [ 141.

Conclusion. - In the temperature range 490-620 K, catalytic oxidation over a

supported silver catalyst is a triangular reaction network. In the investigated par-
tial pressure range (PC2H4= 0.002-0.03 bar and Po,= 0.05-0.2 bar) the partial oxi-
dation to ethylene oxide and the total oxidation to carbon dioxide and water show
the same partial pressure dependence and the apparent activation energies are al-
most of the same magnitude. The ethylene oxide oxidation rate is first order with
respect to ethylene oxide. At low ethylene conversion (XE<O.l) lthe selectivity ob-
tained with respect to ethylene oxide is about 0.55.
We gratefully acknowledge that the present work waspart of a project supported by the Swiss
National Science Foundation.

REFERENCES

[ 11 T.E. Leforf, French patent 729, 952 (193 1).

[2] H. H. Voge & Ch.R. Adams, Adv. Catalysis 17, 151 (1967).
[3] C. H. Bamford & C. F. H. Tipper, ‘Chemical Kinetics’ 20, 126, Elsevier 1978.
[4] D.E. Mears, Ind. Eng. chem. Process Des. Dev. 10, 541 (1971).
[5] G.H. Twigg, Trans. Faraday SOC.42, 284,657 (1946).
[6] K. Murray, Austral. J. Sci. Res. A3, 433, (1950).
[7] Wan Shen- Wu, Ind. Eng. Chemistry 45, 234 (1953).
[8] F. Fognani & R. Montarnal, Rev. inst. franc. pktrole 14, 191 (1959).
[9] F. Alfani & J.J. Carberry, Chim. Ind. (Milan) 52, 1192 (1970).
[lo] R.E. Kenson & M. Lapkin, J. phys. Chemistry 74, 1493 (1970).
[I 11 P.D. Klugherz & P. Harriot, AICHE Journal 17, 856 (1971).
[12] P.L. Metcalf& P. Harriot, Ind. Eng. Chemistry(Proc. Res. Dev.) I I , 478 (1972:).
[13] A . Verma & S . Kaliaguine, J. Catalysis 30, 430 (1973).
[14] O.M. Todes & T. I. Adrianova, 2. Fiz. Chim. 27, 1485 (1953).
[ 15) E. L. Force & A . T. Bell, J . Catalysis 38, 440 (1975).